Alexander Soare
Senior AI Research Scientist at Cobot
About
I'm Alexander Soare, a Senior AI Research Scientist at Cobot. My journey has been a bit of a quest, moving from a background in Quantum Physics and Cryogenic Engineering into the world of operations and eventually deep into the heart of Physical AI. I've spent the last several years focusing on how robots interact with the physical world, from building the LeRobot open-source codebase at Hugging Face to developing 'smooth-as-butter' robot policies. I'm passionate about Embodied AI and the belief that humans and robots working side-by-side is the key to moving the world. Currently, I'm focused on scaling our team at Cobot and pushing the boundaries of what's possible with real hardware. I'm always looking to connect with fellow researchers, engineers, and anyone excited about the future of robotics and open-source intelligence.
Networking
What I can offer
- ›Deep expertise in Embodied AI and robotics perception
- ›Insights into open-source robotics development (LeRobot)
- ›Experience scaling tech infrastructure and operational efficiency
- ›Technical advisory for CV and ML startups
Looking for
- ›Test engineers and interns in robotics and AI for Cobot
- ›Engagement with leading paper authors and researchers
- ›Collaborations within the open-source robotics community
Best fit for
Current Interests
Background
Career
Began in Cryogenic Systems Engineering and Quantum Physics before transitioning through operations and product management at Ontruck. Pivoted into specialized AI/ML roles, eventually leading AI at Dextrous Robotics and contributing to open-source at Hugging Face before joining Cobot.
Education
Master’s Degree, Quantum Physics, University of Sydney (2013 – 2015); Bachelor of Science (Advanced) HONS, Quantum Physics, University of Sydney (2009 – 2012)
Achievements
- ›Founding team member of the LeRobot open-source project at Hugging Face
- ›Developed 'Smooth-as-Butter Robot Policies' for Physical Intelligence
- ›1st Place and D-Wave Prize in the 2020 CDL Quantum Hackathon
- ›Doubled operational efficiency and halved late deliveries at Ontruck
- ›Built MVP and tech infrastructure for DriveandJob acquiring 300 drivers in 3 months
Opinions
- LLMs will not be able to drive Physical AI alone
- Question the validity of assumptions in research methods when outcomes are sensitive to variations
- Humans and robots should work side-by-side to 'move the world'
- Value being hands-on with real hardware while staying at the cutting edge of research